1. Field of the Invention
The present invention relates to an electron beam apparatus which inspects and/or reviews defects on a surface of a sample. More particularly, it relates to a low-voltage scanning electron microscope (LVSEM) with charge-up control for inspecting and/or reviewing defects on a surface of a sample made of substantially insulative material such as wafers or masks in semiconductor manufacturing industry.
2. Description of the Prior Art
In semiconductor manufacturing industry, defects can occur on surfaces of masks and wafers during semiconductor fabrication process. These defects impact yield to a great degree. Apparatuses, which use microscopy to inspect and/or review of defects on a surface of a sample, have been employed to monitor semiconductor manufacturing yield. They are always desired to have high spatial resolution and high throughput. Since critical feature dimensions of patterns on wafers and masks shrunk to ranges where the defects thereon can not be detected out by the photon microscopy, the scanning electron microscopy has been widely adopted. Compared with a photon microscope, a scanning electron microscope (SEM) can detect the defects with smaller sizes due to its higher imaging resolution. However, the incident electron beam (typically called as primary electron beam) of the SEM interacts with the materials of the sample surface and consequently builds a radiation damage and a charge-up (or called as charging) on the sample. The charging may darken, blur and distort the image of the SEM.
The only remedy to reduce radiation damage on the sample surface is to use a low-energy or called as low-voltage electron beam. The lower the energy (Landing Energy) of the primary beam landing on the sample surface is, the less the sample is damaged. In an LVSEM the landing energy is typically lower than 5 keV. Charge-up built on the sample surface results from the difference between the number of incident electrons and the number of emission electrons. FIG. 1, which is cited from the paper in Handbook of Charged Particle Optics, edited by Jon Orloff (CRC Press, Boca Raton, N.Y., 1997), pp. 373-379 by Michael T. Postek, shows the behavior of the total electrons emitted from a sample per unit primary electron with respect to the landing energy, or simply called as total electron yield σ. An amount of net electrical charges will appear on the sample if the yield σis not equal to 1, and only two landing energy values E1 and E2 correspond to the unity yield. E1 and E2 change with material composition of the sample. Typically, the first energy E1 is below 100 eV and the second energy E2 is about 2˜3 keV. Because the inspection or review of defects on wafers or masks is required nondestructive in semiconductor manufacturing, the sample is typically scanned by a primary electron beam having an energy within 0.2˜2.5 keV. For this kind application case, a positive charge-up occurs on the sample surface due to the yield σ larger than 1.
On the one hand, the positive charges form a potential barrier. Some secondary electrons (SE) with low emission energies will be reflected back to the sample surface by the potential barrier. The return of the secondary electrons in turn limits the increase in the sample surface potential but directly results in a reduction in the SE detection. On the other hand, the positive charges change the landing energy of the primary electron beam and the local electrostatic field, thereby defocusing and distorting the primary electron beam. As a result, an SE image of the SEM will be darkened, blurred and/or distorted partially or even all over.
Actual charging situation depends on sample conductivity, dwell time and current (probe current) of the primary electron beam on sample surface as well as total electron yield. Several methods are proposed or employed to control charging or mitigate charging effect on an SEM image. One method is to use an advanced scanning strategy for acquiring a frame of image. One example is the interlacing scanning, which scans every odd number of lines during the first pass and every even number of lines during the second pass. Each pass time can allow the charges to dissipate to reach a charge balance so that a charging-effect mitigated image can be obtained. Another example is Leap-scanning, which scans an area (or called as Field of view) at first and then leap to scan a non-adjacent area so as to avoid impact of charge-up built on the former area. Another method is to use a frame cycle of recovery scanning after each frame cycle of imaging scanning. The recovery scanning is different from the imaging scanning in many ways such as the landing energy and/or probe current of the primary electron beam and the size of the scanning area. For example, in the frame cycle of the recovery scanning, the landing energy of the primary electron beam is chosen corresponding to an electron yield smaller than one so as to generate negative charges to neutralize the positive charges generated during the frame cycle of the imaging scanning. Disadvantageously, the existence of the frame cycles of recovery scanning reduces the throughput obviously.
However, as the throughput is required higher, a higher primary beam current or a larger field of view becomes necessary. In this case, more positive charges will be built up on the sample surface and can not dissipate quickly during a frame cycle of imaging scanning, thereby reducing the FOV size used in the foregoing methods. In addition, as the resolution is required higher, the impact of charge-up on image quality (such as distortion) becomes not negligible. Therefore, the charges built up during a frame cycle of imaging scanning needs to be eliminated as soon as possible.
Accordingly, a new LVSEM, which can eliminate charge-up soon after appearing thereby being able to provide higher spatial resolution and higher throughput, is needed.